OTC 5736

Wind Turbulent Spectra for Design Consideration of

Offshore Structures
by MK Ochi, U. of Florida, and V.S. Shin, American Bureau of Shipping
Copyright 1988 Offshore Technology Conference
This paper was presented at the 20th Annual OTC in Houston, Texas, May 2-5, 1988. The material Is subject to correction by the author. Permission
to copy is restricted to an abstract of not more than 300 words.
ABSTRACT
This paper reviews currently available infor-
mation on turbulent wind spectra over a seaway.
The results of comparisons of spectral densities
constructed from measured data and those
computed by various spectral formulations show a
significant difference at low frequencies. In
order to provide a reliable prediction technique
throughout the entire frequency range, a spec-
tral formulation is proposed for consideration
in the design of offshore structures.
formulation, if necessary, for design considera-
tion of offshore structures. For this, avail-
able wind specta obtained from wind speed
measurements over a seaway are compiled and com-
pared with spectral density functions computed
by available various spectral formulations. It
is found that there is a significant difference
at low frequencies where measured spectral
densities are substantially greater than those
computed by any of the spectral formulations.
SPECTRAL PRESENTATION OF TURBULENT WINDS
INTRODUCTION
One of the areas in which information is
needed in order to establish more rational
design of offshore structures is the estimation
of unsteady wind fields (fluctuating turbulence)
over the sea necessary to evaluate dynamic
response of the structure in a seaway.
The frequency and the spectral density func-
tion of turbulent winds are generally expressed
in dimensionless form. In this paper, spectral
formulations as well as the spectra obtained
from measured data are all expressed in the fol-
lowing dimensionless form for comparison:
Dimensionless frequency:
Dimensionless spectral density:
f = frequency in cps
(2)
(1)
height above sea level in
meters
mean wind speed at height z in
m/sec
density function in
m /sec
z =
S(f) =
where
Currently available turbulent wind spectral
formulations have been developed based primarily
on wind data measured on land in order to pro-
vide information for vibration problems of
buildings. It has been said that spectral den-
sity of turbulent wind over a seaway appears to
be greater at lower frequencies than that
computed by available spectral formulations
(Bush and Panofsky 1968, Simiu and Scanlan 1978,
among others). If this is the case, when a sub-
stantial amount of wind energy exists at a fre-
quency close to the natural response frequency
of the structure, the dynamic effect of turbu-
lent winds on the structure's response can
become a serious problem.
u* = shear velocity in m/sec.
The purpose of this study is to review cur-
rently available information on turbulent wind
spectra over a seaway, and derive a spectral
References and illustrations at end of paper.
The following formulae
U
z
' and friction velocity,
dimensionless presentation
sity function:
for mean wind speed,
are used for the
of the spectral den-
461
2
Wind Turbulent Spectra for Design Consideration of Offshore Structures OTC 5736
U
10
'" mean wind speed at 10 m height
in m!sec,
where
U
z
U
10
+ 2.5 In(z/10)
u* '" {CIa U
10
'
(3)
(4)
TURBULENT WIND SPECTRAL FORMULATIONS
(a) Davenport spectral Formulation
The Davenport spectrum (Davenport 1961) is
given in dimensionless form as
C
10
'" surface drag coefficient
evaluated from wind
measurements at 10 m height.
(5)
The surface dra/I coefficient, C
10
, depends
on mean wind speed, U
10
The results of several
studies on this subject are summarized and shown
in Figure 1. Among others, Wu' s results are
based on data obtained for a wide range of wind
velocities over a seaway; hence, his result
shown in Figure 1 is used in the present study.
WIND SPECTRA OBTAINED FROM MEASUREMENTS OVER A
SEAWAY
where x '" 1,200 f/U
10

As can be seen in the above equation, the Daven-

port spectrum is not a function of height above
the ground (or sea level); instead, it is a
function of the mean wind speed at 10 m height.
(b) Harris Spectral Formulation
The original form of the Harris spectrum is
given in the following form (Harris 1983):
Several turbulent wind spectra have been ob-
tained from wind speed measured over a seaway at
various geographical locations. These are
listed in Table 1 and the spectral density func-
tions are compiled in Figure 2. It should be
noted that some publications give data points
only which are usually scattered over an un-
stable range at low frequencies. The average
curves are drawn through these data points for
the present study and they are shown in Figure
2.
L'
mean square of turbulent wind
fluctuations
characteristic length
(6)
The original form of the Hino spectrum is
given by (Hino 1971),
L
u
'" integral length scale
The characteristic length L' is chosen such
that the spectral density function peaks at
x '" Vickery et al. (1985) state that the
integral length scales, used to fit the Harris'
formulation to the full scale data, range from
60 to 400 m and the mean value of L
u
'" 180 m is
used to match the high frequency portion of the
spectrum. They also suggest the ratio cr!u* '"
2.45 from the observed data. Thus, the Harris
spectral formulation can be written as
where
The spectra shown in the figure are obtained
from wind speeds of various severity, and from
measurements at various heights above the water
surface and at different geographical locations.
Nevertheless, there is little scatter in the
magnitude of the dimensionless spectral density
functions except at very low frequencies.
The upper and lower bound curves are given
for Eidsvik' s wind spectrum. Eidsvik carried
out an extensive analysis of turbulent wind
spectra on a total of 3,660 wind speed time his-
tories covering a mean wind speed range from 2
to 36 m/sec at 110 m height above the water sur-
face. He classified the data into 15 wind speed
groups and generated 15 spectra in sequence of
increasing wind severity. Howeve r, only the
largest and smallest wind spectra are shown in
dimensionless form in Figure 2. For presenting
Eidsvik's data in dimensionless form, the value
of C
10
is maintained constant, 1.5 x 10-
3
, for
all wind speeds, since this value was estimated
through his measurements.
21.97x
S(f*) '" 2 2 5/6
x )
x '" (1.09)(180)/U
z
'
(c) Hino spectral Formulation
(7)
It is of interest to note that Eidsvik' s
upper and lower bound spectra encompass most of
the other data obtained elsewhere. If we draw
the average curve of Eidsvik' s upper and lower-
bound, then it represents approximately the
average of all measured spectra presented in
Figure 2.
462
(8)
mean square of wind turbulent
fluctuations
x '"
a '" exponent of the power law governing
the profile of mean wind speed.
OTC 5736 M.K. Ochi and Y.S. Shin
3
By using the ratio cr/u* s 2.45 given in con-
nection with the Harris spectrum. we may write
the Hino spectrum as follows:
(9)
I
for f* < f
m
for f
m
<f* < f
s
for f* >f
s
(12)
where.
(d) Kaimal spectral Formulation
4 L 13
u
a
1
== --z--
f x s where
(S.34x10
3
) u; (z/10)1-4a
3 -4
a U
lO
Computations are carried out by choosing a
0.16 in the present study.
Kaimal et al. (1972) develop a spectral for-
mulation which includes the effect of dissipa-
tion rate for turbulent energy as a function of
dimensionless height. denoted by z/L. Here. L
is the Monin-Obukhov length which is a function
of shear velocity as well as surface heat flux.
L is negative with upward heat flux (which
depends on the difference between surface tem-
perature and the temperature aloft). L < 0 is
unstable and can be observed in the daytime. The
spectrum is given by.
f
f(6 f -f )2 +:i{2
1
f
2
_f
2
)+2f (f -f )+f (f -2f )R.nf
m s m s m s m ssm m
a2 '" -2 b
2
f
m
2 a
1
f
m
d
1
"'?" [-2- - 13
1
m
( 10)
13 f - b f2
c
2
s 1 - a
2
s 2 s
where
1+0.5 Iz/L1
2/3
1+2.5 Iz/L1
3/S
for -2 ~ z/L ~ 0
for 0 ~ z/L ~ 2
L
u
s integral scale length
f3 '" c r 2 / u ~ s 6.0
f
m
'" dimensionless frequency at which f S(f) is
maximum
z/L s 0 is called the stable condition.
(e) Kareem Spectral Formulation
dimensionless frequency equivalent to the lower
bound of the intertial subrange.
The Simiu and Leigh's spectral formulation
is developed taking into consideration the wind
energy over a seaway in the low frequency range.
In particular. f S(f) becomes maximum at f s f
m

and the spectrum has a finite value at f s 0

given by
Kareem (1985) proposes the following formulae
from analysis of available wind data and
spectral formulations:
(11)
S(O) s 4(cr
2
/u ) L
z u
(13)
In -the derivation of the spectral formulation.
Kareem specifically considers the spectral den-
sity obtained at low frequencies as well as the
agreement between the variance evaluated by
integrating the spectrum and that obtained from
measured data.
From analysis of many full scale measurements.
Simiu and Leigh suggest that the magnitude of L
u
ranges from 100 to 240 m for winds at 20-60 m
above the sea surface. L
u
is taken as 180 m in
the present study.
(f) Simiu and Leigh Spectral Formulation
Simiu and Leigh (1983) suggest the following
empirical spectral formulation for design of
offshore structures:
Figure 3 shows all these spectral formula-
tions. Included also in the figure for compari-
son is the average of all measured spectra shown
in Figure 2. It should be noted that. as stated
in the Introduction. many spectral formulations
presented here are developed based on data
measured primarily over land in an attempt to
formulate turbulent wind energy in the
relatively high frequency range. say f* > 0.1.
However. extertsiort of the formulatiorts is made
here to see if they are still valid in the low
463
4 Wind Turbulent Spectra for Design Consideration of Offshore Structures OTC 5736
frequency range for design consideration of off-
shore structures.
CONCLUSIONS
It can be seen in Figure 3 that there is no
appreciable difference between measured spectra
and those computed by spectral formulations for
dimensionless frequency f* greater than 0.03.
However, a significant difference can be seen
for f* less than 0.03 where all measured
spectral densities are greater than those
computed by any of the spectral formulations.
This paper reviews currently available infor-
mation on turbulent wind spectra over a seaway
and presents a spectral formulation for design
consideration of offshore structures.
From comparison between wind spectra obtained
from data measured over a seaway and those
computed by available various spectral formula-
tions, the following conclusions may be drawn:
Frequencies and periods corresponding to the
dimensionless frequency f* = 0.03 evaluated for
mean wind speeds of 40, 60, and 80 knots at 30 m
height above the sea surface (approximate loca-
tion of the center of gravity for wind force of
offshore structures) are tabulated in Table 2.
It can be seen in the table that the dimen-
sionless frequency f* = 0.03 for a mean wind
speed of 40 knots corresponds to a natural
response period of 48.6 sec. Thus, we may state
that for a mean wind speed of 40 knots extreme
care must be taken in applying the currently
available wind spectral formulations for evalu-
ating motions (surge and sway motions, for exam-
ple) of floating units whose natural motion
periods are greater than 48.6 sec. Similarly,
for structures whose natural motion periods are
greater than 32.4 sec. and 24.3 sec. care must
be taken for mean wind speed of 60 knots and 80
knots, respectively.
(a) There is no appreciable difference between
measured spectra and those computed by spectral
formulations for dimensionless frequencies f*
greater than 0.03.
(b) A significant difference between measured
and computed spectra exists for f* less than
0.03 where all measured spectral densities are
greater than those computed by any of the
spectral formulations. This result is not sur-
prising since currently available turbulent wind
spectral formulations are developed based
primarily on wind data measured on land.
Based on the above findings, the average wind
spectra obtained from measured data over a sea-
way is obtained and presented in the mathemati-
cal formulation given in Eq. (14) for design
consideration of offshore structures.
ACKNOWLEGEMENTS
PROPOSED SPECTRAL FORMULATION FOR TURBULENT
WIND OVER A SEAWAY
For design consideration of offshore struc-
tures, the average wind spectra obtained from
measured data over a seaway is expressed in the
following mathematical formulation as a function
of dimensionless frequency:
The study presented in this paper was carried
out under the Joint Industry Study Project on
Mobile Offshore Drilling Unit Stability admini-
stered by the American Bureau of Shipping. The
authors wish to express their appreciation to
the participants of the Project for their
interest and support of this work and their per-
mission for publication of the results.
REFERENCES
A comparison between the mathematical formu-
lation given in Eq. (14) and the average of
measured wind spectra is shown in Figure 4.
Hino, M.: "Spectrum of Gusty Wind", proc.
3rd Conf. on Wind Effect on Buildings and
Structures, Vol. 1, pp. 69-77, 1971.
Garatt, J.R.: "Review of Drag Coeffi-
cients over Oceans and Continents", Month.
Weather Rev., Vol. 105, pp. 915-929, 1977.
Harris, R.I.: "The Nature of the Wind,
The Modern Design of Wind-Sensitive Struc-
tures", Const. Ind. Res. & Inf. Assn.,
London, pp. 29-55, 1971.
Bush, N.E. and Panofsky, H.A.: "Recent
Spectra of Atmospheric Turbulence", Quart.
Journal Roy. Meteorol. Soc., Vol. 94, pp.
132-148, 1968.
Davenport, A.G.: "The Spectrum of
Horizontal Gustiness Near the Ground in
High Winds", Quart. Journal Roy. Meteorol.
Soc., Vol. 87, pp. 194-211, 1961.
(5)
(2)
(4)
(1)
(3)
(14)
for 0.003 ~ f* ~ 0.1
420 f 0.70
*
(1+*0.35)11.5
838 f*
It is assumed in Eq. (14) that S(f*) =583 f*
for f* < 0.003. This assumption is introduced
because when S(f*) is converted to a dimension-
less spectral density S(f), it becomes infinite
at f = O. Since no information is available on
turbulent wind energy at very low frequencies
(approximately f* < 0.003), the spectral densi-
ties for f* less than 0.003 are modified such
that the magnitude of the density will reduce
linearly to zero at f = O.
464
OTe 5736 M.K. Ochi and Y.S. Shin 5
Smith, S.D.: "Thrust-Anemometer Measure-
ments of Wind-Velocity Spectra and
Reynolds Stress over a Coastal Inlet",
Journal Mar. Res., Vol. 25, pp. 239-262,
1967.
Takeda, A.: "Effects of Water Wave
Motions on Spectral Characteristics of
Wind Fluctuations in the Marine Atmo-
spheric Surface Layer", Journal Meteorol.
Soc. Japan, Vol. 59, No.4, pp. 487-508,
1981.
(6) Kaimal, J.C. et al.: "Spectral Character-
istics of Surface Layer Turbulence",
Quart. Journal Roy. Meteorol. Soc., Vol.
98, pp. 563-589, 1972.
(7) Kareem, A.: "Structure of Wind Field over
the Ocean", Proc. Int'l. Workshop on Off-
shore Winds and Icing, pp. 225-235, 1985.
(8) Large, W.G. and Pond, S.: "Open Ocean
Momentum Flux Measurements in Moderate to
Strong Winds", Journal Phy. Oceano., Vol.
11, pp. 324-336, 1981.
(9) Miller, B.I.: "A Study of the Filling of
Hurricane Donna over Land", Month. Weather
Rev., Vol. 92, pp. 389-406, 1964.
(15)
(16)
(17)
Smith, S.D.: Wind Stress
over the Ocean in Gale
Journal Phy. Oceanography,
709-726, 1980.
and Heat Flux
Force Winds" ,
Vol. 10, pp.
(10) Miyake, M. et al.: "Spectra and Cospectra
of Turbulence over Water", Quart. J. Roy.
Meteorol. Soc., Vol. 96, pp. 138-143,
1970.
(11) Pond, S. et al.: "Measurements of the
Turbulent Fluxes of Momentum, M:>isture,
and Sensible Heat over the Ocean", Journal
Atmosph. Sci., Vol. 28, pp. 901-917, 1971.
(12) Schmitt, K.F. et ale: "Structure of
Marine Surface Layer Turbulence", Journal
Atmosph. Sci., Vol. 36, pp. 602-618, 1979.
(13) Simiu, E. and Scanlan, R.H.: "Wind
Effects on Structures", John Wiley &Sons,
Inc. New York, 1978.
(14) Simiu, E. and Leigh; S.D.: "Turbulent
Wind Effects on Tension Leg Platform
Surge" , National Bur. Stand. , Building
Sci. Sere 151, 1983.
(18)
(19)
(20)
Vickery, P.J. et al.: "Gust Factor and
Turbulence Measured at 50 m over the Sea
Off Sable Island", Proc. Int'l. Workshop
on Offshore Winds and Icing, pp. 196-213,
1985.
Weiler, H.S. and Burling, R.W.: "Direct
Measurements of Stress and Spectra of Tur-
bulence in the Boundary Layer over the
Sea", Journal Atmosph. Sci., pp. 653-664,
1967.
Wu, J.: "Wind-Stress Coefficients over
Sea Surface from Breeze to Hurricane",
Journal Geophy. Res., Vol. 87, pp. 9704-
9706, 1982.
Table 1 Measurements of turbulent winds over the sea
Location
Height above Mean
sea surface Wind speed
in m. in m/sec. Table 2 Frequency and period corresponding
to f* = 0.03 for z = 30 meters
Eidsvik (1985) Norwegian sea 110 2-36
Schmitt et al.(1979) North Pacific 29 5.5-11.8 Mean wind speed at 30 m height
above sea level (in kts)
Miyake et al.(1970)
400 m offshore
1.4-4.5 3.8-8.6
60 80
Vancouver, Canada 40
Takeda (1981)
1 km offshore
0.4-5.1 <7 Frequency (Hz) 0.021 0.031 0.041
Kanagawa, Japan
Barbados
3.9-7.2
Period (Sec.) 48.6 32.4 24.3
Pond et al.(1971)
Caribbean
8
Weiler &Burling(1967)
400 m offshore
1. 7-2. 7 1.4-10.0
Vancouver, Canada
Smith (1967)
Nova Scotia
1. 6-4. 2 6-22
Canada
465
6
Wind Turbulent Spectra for Design Consideration of Offshore Structures OTC 5736
60 50
100KTS
40
70KTS
30 20 10
3
10
~
~
//
/-,vf' /
/.1"/
j,'"
/ ~
V/"
j,'
V ~ ~
V ~
/
" f t ~
... :;-
.,H ... ",
/ ~ ----
GARRATT (1977)
/.# ","
-:
" ~
MILLER (1964)
/ ~
... " ...
--WU (1964)
~ ~ ~ '
---SMITH (1980)
-
~ /
-----LARGE AND POND
(1981 )
I I
o
o
x
5
0
4
0-
~
Z
LU
<3 3
u::
ll.
LU
o
()
~ 2
o
Cl
z
s: 1
WIND SPEED 0101N M/SEC
Figure I Wind drag coefficient, C
IO
' as a function of
mean wind speed, U
IO
0.4 0.60.81.0 0.002 0004CXXX) CDI 002 004 006 0.1 0.2
DIMENSIONLESS FREQUENCY f*=fz/O
z
- ~ ~ ~
/ '"
~
- ,
"
""""
"-"
~
EIDSVIK
' .....
~ ---------SCHMITT ET. AL.
........ "
'"
MIYAKE
r......
---TAKEDA
---POND
----WEILER & BURLING
-----SMITH
8
1
001
~
~ 20
b
~ 10
~ 0.8
..J 0.6
~ 04
ffi
~
o
0.2
10.0
80
60
J: 40
~
(/)
Figure 2 Comparison of turbulent wind spectra obtained from
data measured over a seaway
466
OTC 5736 _ .M.l<. OcllLandY.S ..Shin
7
04 06081.0
-
KAIMAL
--- '- -
_.
--- DAVENPORT
---KAREEM
---- HARRIS
Ave. of Measured Spec!.
----- HINO
J
-------- SIMIU
.----
- -

--
;;7'i

t>
-

---
,......
1-- "
""-.
'/"'"

"
I,/,
, /
./
\

/'
/; -'
/.-----
f.--";" , r-::

"
I-.....
,

/'
,
V ,
-

, ,
1'-
/
,/,-
,
,/'
V
,/


"
...
./
, ,
./ ,
,
...
/
'/
/
1/
--f---.

,-

,
/
/
[/
"

,
"V' /
/'J
0.1
0001 0002 0004c:x:x:6001 002 004006 01 0.2
DIMENSIONLESS FREQUENCY f*=fz/O
z

20
td
a..
(f) 10
0.8
0.6
z
04
z
w

5 0.2
40
(f)
100
80
6.0
Figure 3 Comparison of various wind spectral formulations
583 f* for 0.003
-
I
420
'7 035-.115 for 0.003:s; f*:s;0.1
41--+----l-------1f----+--I S(f *)= (1 +L' ) . -
::2: / 838f*
::> for 0.1
2 /,:::::-_ -/-
1 /' Ii'-",_
CI) 1----+-_-+ Average of Measured Spectra

0.6
0.4 '"
ill
::2:
Cl
0.2 f---l----f---+---+---l----1---f---i----+---l--t------i
0.002 0.004 0.01 0.02 0.04 0.06 0.1 0.2
DIMENSIONLESS FREQUENCY f *= fzlO
z
0.4 0.6 1.0
Figure 4 Mathematical presentation of average of measured wind
spectra over a seaway
467